PLAYER

Introduction

This repository contains the code submitted to COLM for PLAYER: Enhancing LLM-based Multi-Agent Communication and Interaction in Murder Mystery Games.*

Requirements

• Python >= 3.10
• pandas >= 1.5.3
• faiss-cpu >= 1.7.4
• transformers >= 4.38.2

Dataset

The bilingual dataset is housed within the ./chinese and ./english folders. For detailed information, please refer to the README.me files located in these folders, available in both Chinese and English versions, respectively.

Gameplay and Evaluation

To run our method on the Chinese dataset, please use the following:

Python main.py --script_name "孤舟萤（6人）" --output_root_path "./log_cn"

Here, script_name is the name of the script you want to run.

To run our method on the English dataset, please use the following:

Python main.py --script_name "Solitary Boat Firefly (6 people)" --output_root_path "./log_en" --is_english 1

The default model is the GPT-3.5 16k version. However, we also offer code support for using models such as LLaMA2 (7B, 13B, 70B) and Gemma7B. Please include the --model_type parameter and add the path to your model in the main.py file.

Gameplay and Evaluation log

The log files are stored in the ./log_cn and ./log_en folders. The log files contain the following information:

• *.xlsx: The log file contains the dialogue history and the generated responses.
• evaluation.xlsx: The log file contains the evaluation results of the generated responses.